ChatGPT Scored 72% In Clinical Decision Accuracy, MGB Study Shows

According to an MGB research, ChatGPT has a 72% clinical judgment accuracy rating.According to new research from Mass General Brigham, the broad language model had inconsistent performance throughout general and emergency care, as well as for all medical specialties, although it struggled with differential diagnosis.Mass General Brigham researchers put ChatGPT to the test to see if AI can handle an entire clinical encounter with a patient, including suggesting a diagnostic workup, choosing a course of action, and making a final diagnosis. They discovered that the large language model had "impressive accuracy" despite limitations, such as potential hallucinations.